Field of the Art
The disclosure relates to the field of image processing and linear feature extraction, and more particularly to the field of extracting three-dimensional hydrology vector models from remotely-sensed imagery.
Discussion of the State of the Art
In the art of linear feature extraction, ROADTRACKER™ and similar tools enable automated bulk extraction and semi-automated point-to-point extraction of two-dimensional linear feature vectors from remotely-sensed imagery. Extractions by these tools are image-based, meaning image content automatically influences the trajectories of the extracted vectors. In semi-automated extraction, the image raster is displayed in a viewer and extraction is partially guided by user mouse clicks placed along a desired linear feature. Tools like ROADTRACKER™ can be used to perform two-dimensional extraction of, among other things, single-line drainages (the centerline vectors of narrow streams), double-line drainages (the two side vectors of rivers), and the boundaries of water bodies, and includes automatic smoothing of the extracted vectors and automatic topology cleaning of the vectors (elimination of gaps (under-shoots) and dangles (over-shoots) where vectors are intended to be perfectly incident to one another.) In this manner, a two-dimensional hydrology vector model for a raster image may be constructed.
What is needed is an automated way to convert a two-dimensional hydrology vector model into a corresponding three-dimensional hydrology vector model where the latter is geometrically consistent with the physics of real-world hydrology, i.e., lake boundaries are level, corresponding points on opposite sides of a river are at the same altitude, and water flow is monotone decreasing in altitude.